view Snakemake_files/Snakefile_orthofinder_heatmap_upset @ 5:4157f435fa89 draft

Uploaded
author dereeper
date Thu, 30 May 2024 12:11:25 +0000
parents e42d30da7a74
children
line wrap: on
line source

import glob
import os
import shutil

import yaml
configfile: "config.yaml"

SAMPLES = []
with open("config.yaml", "r") as yaml_file:
    genome_data = yaml.safe_load(yaml_file)
    for genome_name, genome_info in genome_data["input_genomes"].items():
        name = genome_info["name"]
        SAMPLES.append(name) 



rule final:
 input:
  expand("outputs/genomes/{sample}.fasta", sample=SAMPLES),
  expand("outputs/genomes/{sample}.gff", sample=SAMPLES),
  expand("outputs/genomes/{sample}.pep", sample=SAMPLES),
  "outputs/pav_matrix.tsv",
  "outputs/heatmap.svg.gz",
  "outputs/rarefaction_curves.txt"

rule get_data:
    input:
        "config.yaml"
    output:
        "outputs/genomes/strains.txt",
        expand("outputs/genomes/{sample}.fasta", sample=SAMPLES),
        expand("outputs/genomes/{sample}.gff", sample=SAMPLES),
    run:
        with open(input[0], "r") as yaml_file, open(output[0], "w") as output_file:
            genome_data = yaml.safe_load(yaml_file)
            for genome_name, genome_info in genome_data["input_genomes"].items():
                fasta_file = genome_info["fasta"]
                name = genome_info["name"]
                gff_file = genome_info["gff3"]
                shutil.copy(gff_file, "outputs/genomes/"+str(name)+".gff")
                shutil.copy(fasta_file, "outputs/genomes/"+str(name)+".fasta")
                output_file.write(f"{name}\t{name}\n")

rule gffread:
    input:
        fasta = "outputs/genomes/{sample}.fasta",
        gff = "outputs/genomes/{sample}.gff"
    output:
        "outputs/genomes/{sample}.pep"
    shell:
        """
        gffread -y {output} -g {input.fasta} {input.gff}
        """

rule orthofinder:
    input:
        expand("outputs/genomes/{sample}.pep", sample=SAMPLES)
    params:
        identity=80
    output:
        pav="outputs/Orthogroups.tsv"
    shell:
        """
        mkdir -p outputs/proteomes
        sed -i '/^>/! s/\./N/g' outputs/genomes/*.pep
        cp -rf outputs/genomes/*.pep outputs/proteomes
        orthofinder -f outputs/proteomes
        mv outputs/proteomes/OrthoFinder/R*/Orthogroups/Orthogroups.tsv outputs/Orthogroups.tsv
        """

rule convert_matrix:
    input:
        pav="outputs/Orthogroups.tsv"
    output:
        "outputs/pav_matrix.tsv"
    shell:
        """
        perl $PANEX_PATH/Perl/ConvertOrthofinderMatrix.pl outputs/genomes {input} {output} outputs/genomes/strains.txt
        """

rule heatmap_upset:
    input:
        pav="outputs/pav_matrix.tsv"
    output:
        heatmap="outputs/heatmap.svg.gz",
        html="outputs/heatmap.svg.heatmap_plotly.html",
        upsetr="outputs/upsetr.svg",
        binpav="outputs/heatmap.svg.pangenes_01matrix.txt"
    shell:
        """
        perl $PANEX_PATH/Perl/GenerateHeatmapFromPAV.pl {input.pav} outputs/heatmap.svg
        mv outputs/heatmap.svg.upsetr.svg {output.upsetr}
        """

rule micropan:
    input:
        binpav="outputs/heatmap.svg.pangenes_01matrix.txt"
    output:
        txt="outputs/rarefaction_curves.txt",
        pdf="outputs/rarefaction_curves.pdf",
        svg="outputs/rarefaction_curves.svg",
        heaps="outputs/heaps.tsv"
    shell:
        """
        Rscript $PANEX_PATH/R/micropan_rarefaction.R -f {input.binpav} -p {output.pdf} -a {output.heaps} -o {output.txt}
        pdf2svg {output.pdf} {output.svg}
        """